ShadowGraph
Start free trial
Upstream signals

When several companies on one provider report trouble at once, the provider is the story.

We read the official status pages of — companies every ten minutes and know which cloud, CDN and DNS provider each one runs on. When companies that share a provider declare incidents within the same hour, and that is very unlikely to be coincidence, we raise a signal, even if the provider has said nothing.

Last check
—
Status pages read
—
Shared-provider groups
—
Rule
—

Loading the latest results…

No active upstream signal

No group of companies that share a provider is declaring incidents together right now. Checked —.

How a signal is raised

Three steps, every ten minutes. The rule is fixed and versioned; changing it means a new version and a public changelog entry.
  1. Read what companies sayOfficial incident declarations from — machine-readable status pages: GitHub, Vercel, Twilio, OpenAI, Datadog and many more.
  2. Group them by shared providerOur dependency map knows each company's cloud region, hosting provider, CDN and DNS provider. Companies that share one form a group.
  3. Ask: could this be chance?Each company has its own normal incident rate. If at least two companies of a group declare within one hour and the chance of that by coincidence is below 1 in 1,000, we raise a signal.

Track record

Measured over the last — days by running the same rule over the full public history. Evidence is checked against the providers' own records (AWS Health Dashboard, Google Cloud, Cloudflare, Akamai, Vercel) and the companies' own words.
Signals per week—— signals in — days. Rare by design.
Backed by evidence——
Confirmed by provider—Providers often say nothing.
Provider incidents caught—Of provider incidents where at least two of our companies declared too.

Honest reading: signals are rare and usually right about the provider, but most provider incidents do not show up as simultaneous declarations of the companies we follow. Full method, sensitivity analysis and every signal: Signal track record.

Recent signals

Get the signal for your own vendors

Stack Alerts sends provider incidents, early warnings and upstream signals for the services you depend on to email, Slack or Teams.

Start the 14-day trial